Glucose Monitor

Overview

The Glucose Monitor is an advanced breath-based glucose monitoring system that utilizes a Raspberry Pi, sensors, machine learning, and a graphical user interface (GUI) to estimate glucose levels non-invasively. This project is ideal for demonstrating skills in hardware integration, machine learning, and software development.

The system predicts glucose levels based on acetone levels detected in breath using a Random Forest regression model. It features real-time monitoring, data logging, and visualization capabilities.

---

Features

• Sensor Integration: Reads acetone levels using the MQ-138 sensor connected to a Raspberry Pi via MCP3008 ADC.
• Machine Learning: Predicts glucose levels using a trained Random Forest regression model.
• Interactive GUI: Displays real-time glucose level estimates with a Tkinter-based interface.
• Data Logging: Logs sensor readings and predictions to a CSV file for later analysis.
• Expandable Architecture: Modular codebase for easy extension and scalability.
• Real-Time Updates: Continuously updates glucose level predictions every 2 seconds.

---

Project Structure

Glucose_Monitor/ ┣━ main.py # Main program entry point ┣━ gui.py # GUI-related code ┣━ sensor.py # Sensor reading logic ┣━ model.py # Machine learning model training and prediction ┣━ utils.py # Utility functions (data logging, alerts) ┣━ data/ ┃ ┗━ glucose_log.csv # Logs of glucose readings ┣━ requirements.txt # Python dependencies ┗━ README.md # Project documentation

---

Requirements

This project requires the following hardware and software:

Hardware:

• Raspberry Pi (any model with GPIO support)
• MQ-138 gas sensor (for acetone detection)
• MCP3008 ADC (Analog-to-Digital Converter)
• Jumper wires and breadboard

Software:

• Python 3.7 or higher
• Libraries (see requirements.txt)
  • numpy
  • pandas
  • scikit-learn
  • matplotlib
  • RPi.GPIO
  • tkinter

---

Installation

1. Clone this repository:

git clone https://github.com/yourusername/Glucose_Monitor.git cd Glucose_Monitor

2. Install the required Python libraries:

pip install -r requirements.txt

3. Set up your Raspberry Pi hardware:

• Connect the MQ-138 sensor to the MCP3008 ADC.
• Wire the MCP3008 to your Raspberry Pi GPIO pins as specified in sensor.py.

4. Run the program:

python main.py

---

Usage

Real-Time Monitoring:

1. Launch the program by running main.py.
2. The GUI will display real-time glucose level predictions based on acetone readings from the sensor.
3. The system updates every 2 seconds.

Data Logging:

• All sensor readings and predicted glucose levels are automatically saved to data/glucose_log.csv for future analysis.

Alerts:

• You can extend the project to send alerts via email or SMS if glucose levels exceed safe thresholds (see utils.py for customization).

---

How It Works

1. Sensor Reading:

• The MQ-138 sensor measures acetone levels in breath.
• The Raspberry Pi reads these values via MCP3008 ADC.

2. Machine Learning Model:

• A Random Forest regression model predicts glucose levels based on acetone values.
• The model is trained on simulated data but can be updated with real-world data.

3. GUI:

• A Tkinter-based GUI displays real-time glucose estimates.
• Users can monitor their predicted glucose levels interactively.

4. Data Logging:

• Sensor readings and predictions are logged into a CSV file for analysis and visualization.

---

Future Enhancements

Here are some potential ways to expand this project:

1. Cloud Integration:

• Send data to a cloud platform (e.g., AWS IoT or Firebase) for remote monitoring.

2. Mobile App:

• Create an Android/iOS app to display real-time glucose data.

3. Improved Machine Learning:

• Use more advanced models like Gradient Boosting or Neural Networks.

4. Real Data Integration:

• Train the model with real-world breath-based glucose datasets.

5. Alerts & Notifications:

• Integrate Twilio or an email API to send notifications when glucose levels are abnormal.

---

License

This project is open-source under the MIT License.

---

Acknowledgments

Special thanks to open-source contributors and developers who maintain libraries like Scikit-learn, Matplotlib, and RPi.GPIO, which made this project possible.